Reliability of social survey data on noise effects

The results of household interviews provide essential data for the formulation of noise control policies. The few previous studies which have investigated the reliability of such data have yielded low test–retest correlations for measures of annoyance at road traffic noise. This paper complements those previous studies by investigating four related topics: the equivalence of three commonly used annoyance scales; the test–retest reliability of two such scales, for aircraft and overall noise as well as for road traffic noise; the repeatability of measures of average response, such as percent highly annoyed; and the effect on reliability estimates of the range of noise levels included in a survey. The data are from a sample of 212 persons in the vicinity of Toronto International Airport, who were interviewed one year apart, in 1978 and 1979. The sample was stratified by both aircraft and road traffic noise levels. The results show that three commonly used annoyance scales are acceptably equivalent (a composite index comprised of the three has reliability of α = 0.85), and that the effect on the reliability of the range of noise levels included appears to be minimal. The test–retest results are similar to previous findings, suggesting that instability of attitudes toward noise, as well as other sources of measurement error, affect such estimates, and reduce retest estimate of reliability. Much better retest reliability (r≳0.90) is found for measures of average group response such as the percent highly annoyed than for individual level response measures.